New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
use openblas instead of gslcblas #5063
Conversation
please test |
The tests are being triggered in jenkins. |
A new Pull Request was created by @smuzaffar (Malik Shahzad Muzaffar) for branch IB/CMSSW_11_0_X/gcc700. @cmsbuild, @smuzaffar, @gudrutis, @mrodozov can you please review it and eventually sign? Thanks. |
-1 Tested at: a849fb5 You can see the results of the tests here: I found follow errors while testing this PR Failed tests: UnitTests
I found errors in the following unit tests: ---> test DetectorDescriptionRegressionTestDOMCount had ERRORS |
Comparison job queued. |
Comparison is ready Comparison Summary:
|
There is one small difference in a jet b-tag discriminant variable
It's hard to say which one (I only plot by the vector index); it may even be one of the mxnet-based cases. @smuzaffar please clarify if this change (alone, or at least without much else related) can be made available in the DEVEL branch IBs. |
@slava77 , DEVEL IBs are broken due to tensorflow update. I can include it in gcc8 or root 618 IBs |
@slava77 Is this difference in AK4 jets or AK8 jets? Only AK8 jets have MXNet-based taggers. |
in AK8 |
OK I will take a look. |
@slava77 I confirm that there are some small numerical changes on the MXNet-based tagger outputs after changing the BLAS library. The differences are small enough so they should not have any impacts on the physics, but will unfortunately affect the reproducibility of existing NanoAOD workflows. |
@hqucms This sounds "good enough" for 11_X then. |
+externals |
This pull request is fully signed and it will be integrated in one of the next IB/CMSSW_11_0_X/gcc700 IBs (but tests are reportedly failing). This pull request will now be reviewed by the release team before it's merged. @davidlange6, @slava77, @smuzaffar, @fabiocos (and backports should be raised in the release meeting by the corresponding L2) |
@smuzaffar @slava77 @hqucms how urgent is this/ i have already started the build, of course we can integrate, launch an IB and restart in parallel, if this is felt as urgent |
In the context of potential backports and the next pre potentially falling into August and slow response from validators (in August), it would be nice to have it in pre3. |
+1 @smuzaffar I would suggest to start a build with package integrated, and restart the 11_0_0_pre3 build with it |
sure @fabiocos , I am triggering an IB now. |
@smuzaffar thanks |
DYNAMIC_ARCH=1
gslcblas
library from gsl toolfile and addopenblas
dependency to make sure that any cmssw package using gsl linksopenblas
instead ofgslcblas
-lopenblas
instead of-lgslcblas
geneva
,thepeg
andherwigpp
specs to link againstopenblas